BMW tests next-gen LiDAR to beat Tesla to Level 3 self-driving cars
BMW tests next-gen LiDAR to beat Tesla to Level 3 self-driving cars
Tesla's autonomous vehicle tech has been perennially stuck at Level 2 self-driving, as BMW and other rivals try to leapfrog to Level 3.
BMW tests next-gen LiDAR to beat Tesla to Level 3 self-driving cars::Tesla's autonomous vehicle tech has been perennially stuck at Level 2 self-driving, as BMW and other rivals try to leapfrog to Level 3.
Tesla's decision to only use cameras and no lidar will bite them in the ass.
*Musk’s. He regularly overrules the Tesla engineers.
This. That cocksucker has such ~a tiny dick~ fragile ego he makes huge decisions without any expertise simply because he says so. Thats how he built the whole “genius” thing around him. Reality of it is that he is an annoying dumbass who thinks he knows it all and anyone in the same room with more than one brain cell is immediately annoyed with him. But he has a lot of money so i guess LeTtEr X cOoL
Came here to say this. Couldn’t be more on point. Using both cameras and LiDAR in tandem will be necessary for true self driving vehicles.
fortunately LIDAR unit costs are going down, so multiple units, fusing their data with regular camera arrays should resolve a very good view, and be good at error-correcting for each other's shortcomings.
Already is.
Humans drive using “cameras” (eyes) and no LiDAR, that’s the assumption Tesla is making — that a supercomputer can drive 10x better than humans using the same type of sensor. Nobody really knows yet if that’s true but I get the logic.
LiDAR also is UV/visible spectrum and is blocked by dust/fog/snow/rain so it doesn’t help much in many driving situations..
You're making an argument against LiDAR with it using UV/visible spectrum, guess what uses visible spectrum to see stuff? Cameras. And they also have an unfortunate downside of not having good dynamic range, so in very bright/low light situations they probably don't work that well either. Teslas aren't even using infrared cameras to see in the dark to my knowledge.
Unsure why you are downvoted, because that is sound logic. I recall hearing on a podcast of I think a former Tesla engineer that having too many sensors potentially makes things less effective since you have to deal with different types of input, and have to crunch more data, etc. etc. Efficient development also means knowing when to cull unnecessary time sinks.
I hate Elon as much as the next guy, but... Well, humans are obviously not perfect drivers, but Tesla clearly believes that in time, with cameras all around the car (already an improvement over human drivers), a good enough AI solution would be able to match or surpass humans.
Currently they seem to be leading the race though even though the competition is using radar and lidar
Edit: Am I wrong?
If buy leading the race you mean the only company to have an actual product available for purchase then yeah.
But the reason they were able to get to market so quickly is because they don't actually have any concerns about it being functional or safe. That's a real boon to them because it helps them move quickly ahead of the competition that do care about those things.
Of course one good argue that an unsafe self-driving system is in fact not a self-driving system and therefore they are not the first to market.
Cameras are better for surveillance of people and can better be sold. Lidar data not.
I hate that the article opens with
Just a decade ago, the concept of self-driving cars might have seemed like something out of a science fiction movie
Ten years ago there was already a ton of competition in self driving car research. They were first legalized on the roads 10 years ago. Tesla autopilot (including it even though it was a scam) was sold 9 years ago. Google spun off its self driving car division as waymo in 2016.
This feels like one of those "bruh Zelda ocarina of time came out 29 years ago, we old" memes
Hell, Mercedes and Bosch were testing it all the way back in 1993:
Here is an alternative Piped link(s): https://piped.video/watch?v=JTnBiTIvGqY
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I'm open-source, check me out at GitHub.
Old Mercedes was insane. They really were the best or nothing
And what happened though, was it unfilled hype or that death caused by that Uber autopilot? Back when waymo was grabbing headlines you would've thought we'd be a lot closer to driver less by now.
Bet even the self driving software of BMW won't use turn signals when they change lanes.
It would be a bug if it did signal 🤣
Turn what?
It's that thing poor people do right before you cut them off
This will probably be under monthly subscription
"The route you selected contains a highway. Please purchase the Highway Driving Pack in addition to your City Driving Pack to reach your destination"
Fekkin hell that sounds like a boring dystopia
Sounds like a pretty bad Black Mirror episode
Hey, Presto! You've invented toll roads!
Almost certainly.
But self-driving also depends on up-to-date mapping data and continually improved algorithms for the autonomous systems to work properly. An ongoing cost to the customer makes the most sense for a service that has operating costs to the service provider.
I mean, does it? Presumably the idea (that Tesla had anyway) is to try and mimic what humans do, and we don't need mapping data to drive "safely" (for a given value of safe). Of course, humans also get lost, but again, the GPS updates is basically free at this point for the mapping help humans need. (Garmin stopped charging yearly long ago, Open Maps and Google Maps and Wayze all are "free").
The goal is than no one owns anything anymore. Every company is after the sweet sweet recurring revenue.
GM has had supercruise under subscription service for forever.
Sample pricing for BMW self-driving add-on feature:
98% accuracy in obstacle avoidance - $299/mo.
85% accuracy in obstacle avoidance - $199/mo.
75% accuracy in obstacle avoidance (lowest legal limit!) - $99/mo.disclaimer: BMW cannot guarantee 100% accuracy in accuracy rates
Another possibility.
Unlimited* crash avoidance instances - $299.99/mo
10 crash avoidance instances - $199.99/mo
5 crash avoidance instances - $99.99/mo
*crash avoidance may be limited during peak hours and times of congestion. After 12 crash avoidance instances, feature may be disabled without notice due to abuse of the system. All sales are final and minimum 5 year contract required. Price may increase at any time without notice
I've always thought that the Tesla craze would fizzle as major car brands start investing in EVs and self driving tech. I'll take a Toyota, Volvo, Honda or BMW over a Tesla anytime.
Sadly Toyota is struggling to make a decent EV years after leading in hybrids. BMW on the other hand has insane efficiency
hopefully this will change soon, there's some nifty shit on the horizon from Toyota, hope they meet expectations. https://www.inverse.com/tech/toyota-electric-car-600-mile-range-ev
My understanding was that the challenge in making the next leap in self driving was not based in hardware (detecting objects with cameras vs LiDAR), but in software. As in, it isn't as difficult to detect the presence of objects as it is to make consistent and safe decisions based on that information.
But using LIDAR, you increase your data's accuracy and dimensionality, giving you more options to play with. It probably won't be a game changer, but it may be better than a camera only system.
That's not necessarily true. What you get is two separate things inputting raw data into a system that both need to be parsed. Sometimes, one won't agree with the other and can cause issues with how the car thinks it should respond.
Nobody has a fully working system at this point, so it's premature to make claims about what hardware is and isn't needed. It may very well be that LIDAR is a requirement, but until somebody figures it out, we're all just speculating.
Self driving cars are great and all, but can we get someone seriously working on alternative fuels? EV is really pretty unsustainable. All the resources going to build batteries that are unrecycleable is a massive waste in my opinion. And the unless something drastic changes, the ranges that are needed for logistics and America aren't going to ultimately fix anything.
If they can create an alternative fuel that is significantly less polluting, or figure out how to make hydrogen less explody, the existing infrastructure worldwide of gas stations can still be efficiently used. And hopefully there will be a to retrofit existing vehicles to use this alternatives.
batteries that are unrecycleable
Is this actually the case?
My understanding is that EV batteries are actually very recyclable, up to 90%. I imagine it's more labor intensive than your conventional lead-acid batteries though.
I mean, shouldn't we be working on both? Just because they're working on one, that doesn't necessarily mean they're not working on the other.
They also are working on alternative fuels in a big way. Japan have made some incredible leaps with hydrogen/ammonia based production and fuels, and solid state batteries are looking to be pretty game changing. The EU also included a huge budget to invest in green fuels research (likely because of automotive companies lobbying for it) so plenty is being done. Even if EVs aren't the best currently, increasing the size of the market for them will continue to create investments in serving those markets more efficiently, so we absolutely should keep investing in both.
I'm not an expert or anything, but I doubt we'll see a price-competitive synthetic fuel in the time it takes for renewables to become the standard. Renewable electricity gets cheaper as more panels and turbines are built, so it makes some economical sense too.
I think we should NOT allow self driving cars.
> but but muh autopilot in planes
No! Planes are inherently safer to fly than it is to drive cars. Planes have much more room and degree of freedom to maneuver. They are also monitored by air traffic control and the pilots are supposed to be highly trained and fit. Cars are restricted to one plane (heh) and any two bit yokel with stroke-diabeetus-fetal alcohol syndrome from Bumfuck Florida can get a license to drive a car. You can't pull up or dive in a car more than once.
> buh buh buh make every car automated
You just reinvented the train.
My favorite part of a train is the part where it drops you off right at your door.